Ultrasmall pH-responsive silicon phthalocyanine micelle for selective photodynamic therapy against tumor

1. Introduction

In order to further enhance the effect of tumor photodynamic therapy (PDT) and reduce the damage to normal tissues, targeted PDT has been proposed through designing photosensitizers that aim at tumor characteristics.^1,2,3,4,5 Compared with inefficient passive targeting dominated by the enhanced permeability and retention (EPR) effect, the active targeting strategies based on tumor-specific molecular recognition have attracted more and more attention. In view of excellent antigen specificity, monoclonal antibodies have become the first choice for tumor targets.^6,7 ASP-1929, a photosensitizer drug composed of cetuximab and water-soluble silicon phthalocyanine, is already in clinical trials.^8,9 In addition, other target molecules such as peptides,^10,11,12,13 folic acid^14,15,16,17 as well as glycosyl^18,19 have also been tried, accompanied by good tumor enrichment and retention. However, the damage of these actively targeted photosensitizers to normal tissues is inevitable on account of their high photosensitivity during blood circulation.

Ideal photosensitizers should maintain low or even no photosensitivity in the blood circulation but be specifically activated at the tumor site. This activatable strategy is considered to be an effective method for solving the deficient selectivity of photosensitizers.^{20,21,22,23,24,25} The tumor microenvironment (TME) exhibits many unique properties such as faintly acidic pH, high expression of matrix metalloproteinase and glutathione, which provides excellent response sites for triggered PDT against tumor.^26,27,28,29 Moreover, high reactive oxygen species (ROS) produced during PDT were also utilized to restore photosensitizer activity at the tumor site.^30,31 Among them, the universality of acidic TME regardless of the tumor types or the developmental stages enables the pH-activatable photosensitizer with extreme potential.

The large amount of lactic acid resulted from the unique glycolysis of tumor cells that transforms TME into a faintly acidic state in sharp contrast with normal weakly alkaline physiological environments. This acidic microenvironment exists in almost all solid tumors, making pH response an attractive broad-spectrum tumor target for selective PDT.^{32,33,34,35,36} Some pH-sensitive units such as amino groups and amino acids have been used to construct pH-activated phthalocyanines with good pH response and anti-tumor effects.^37,38 Nevertheless, poor water solubility forces these phthalocyanine molecules to require vast excipients for aiding injection. Coupling with polymers or silica to form nanoparticles is conducive to improving the solubility of phthalocyanine, while it also introduces some new defects, including low drug loading capacity, complex synthesis, difficult degradation and metabolism, and potential biotoxicity.^{39,40,41,42,43,44,45}

To address the aforementioned limitations, we designed an ultrasmall pH-responsive silicon phthalocyanine nanomicelle (PSN) by attaching amino groups and polyethylene glycol (PEG) on two sides of the silicon phthalocyanine axis, respectively. Covalent bonding enables PSN to possess a high drug loading capacity and prevent the phthalocyanine from leaking during blood circulation. Inherent amphiphilicity promotes PSN to self-assemble into micelles beneficial to drug injection and delivery so that PSN locates tumors through the EPR effect. The photon-induced electron transfer (PET) effect of amino groups will keep PSN low in fluorescence and photosensitivity during blood circulation to minimize the toxicity to normal tissues. Once PSN reaches tumor sites, acidic TME protonates amino groups to reduce the PET effect, thereby restoring the fluorescence and photodynamic activity of PSN. Furthermore, positive charges after protonation can promote the uptake and retention of PSN by tumor cells, which further enhances the photodynamic anti-tumor effect of PSN (Scheme 1). Thus, PSN is a pH-responsive photosensitizer that can specifically kill tumors with negligible damage to normal tissues, and it has broad application prospects for selective PDT against tumor.

2. Experimental Section

2.1. Materials

1,3-diiminoisoindoline and p-hydroxyhydrocinnamic acid were purchased from Tianjin Heowns Biochemical Technology Co., Ltd. Silicon tetrachloride was obtained from Beijing Innochem Technology Co., Ltd. 1,3-Bis(dimethylamino)-2-propanol was acquired from Sigma-Aldrich. Amino-polyethylene glycol 2000 was purchased from Hefei Bomei Biochemical Technology Co., Ltd. 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) was gained from Shanghai Aladdin Bio-Chem Technology Co., Ltd. 1,3-diphenylisobenzofuran (DPBF) was purchased from Ark Pharm, Inc. DCFH-DA was obtained from Jiangsu KeyGEN BioTECH Co., Ltd. CCK-8 cell viability assay kit was purchased from Shanghai Beyotime Biotechnology Co., Ltd. Lysotracker was acquired from Suzhou Meilun Biotechnology Co., Ltd. 4${}^{\prime }$,6-diamidino-2-phenylindole (DAPI) and Apoptosis Detection Kit were gained from Beijing Solarbio Science & Technology Co., Ltd. Dulbecco’s modified Eagle medium (DMEM) and Fetal Bovine Serum (FBS) were purchased from Thermo Fisher Scientific. Other reagents were of analytical grade and were employed without further purification.

2.2. Synthesis of PSN (Fig. S1)

2.2.1. Synthesis of silicon phthalocyanine dichloride

The solution of 1,3-diiminoisoindoline (10g) and silicon tetrachloride (10mL) in quinoline (200mL) was heated to $180^{\circ }$C for 2h. After cooling to room temperature, it was poured into 400mL dimethyl formamide (DMF) followed by suction filtration. The filter residue was repeatedly washed with chloroform, dichloromethane, methanol, acetone, tetrahydrofuran, and benzene until the filtrate was colorless. A pure bright purple solid was obtained as silicon phthalocyanine dichloride (11.2%) after drying.

2.2.2. Synthesis of PEG-p-hydroxyphenylpropionic acid intermediate

p-Hydroxyphenylpropionic acid (83mg) and EDC (105mg) in anhydrous DMF (5mL) were stirred for 10min at ambient temperature. Then, PEG-NH₂ (100mg) was added and the mixture was stirred overnight. The crude product after removing solvent was purified with a silica gel column (eluent: methanol:$\mathrm{\text{dichloromethane}}=1\text{:}15$, 1% glacial acetic acid was added) to obtain the PEG-p-hydroxyphenylpropionic acid intermediate (82.5%).

2.2.3. Synthesis of compound 1

Silicon phthalocyanine dichloride (100mg), sodium hydride (16.4mg) and 1,3-bis(dimethylamino)-2-propanol (95mg) in anhydrous toluene (15ml) were refluxed for 24h. The crude product, after removing solvent, was purified via a basic alumina column with ethanol/chloroform (1:9) as the eluent. ¹H NMR (500MHz, CDCl₃) $\delta$ 9.61 (dd, $J=5.5$, 3.0Hz, 8H), 8.29 (dd, $J=5.7$, 2.8Hz, 8H), 0.48 (s, 24H), $-$0.80 (dd, $J=12.5$, 5.4Hz, 4H), $-$1.56 (dd, $J=12.4$, 5.1Hz, 4H), $-$2.64 to $-$2.79 (m, 2H) (Fig. S3).

2.2.4. Synthesis of PSN (compound 2)

Compound 1 (100mg), sodium hydride (4.8mg) and PEG-p-hydroxyphenylpropionic acid intermediate (240mg) in toluene (15ml) were refluxed overnight. The crude product, after removing solvent, was purified via a basic alumina column with triethylamine/ethyl acetate (1:1000) as the eluent. PSN (23.8%) was obtained by collecting blue-green bands. ¹H NMR (500MHz, CDCl₃) $\delta$ 9.62 (s, 8H), 8.34 (s, 8H), 3.71–3.52 (m, 175H), 2.34–1.98 (m, 11H), 1.71 (dd, $J=18.4$, 10.4Hz, 4H), 0.52 (s, 3H), 0.48 (s, 8H), $-$0.75 (s, 1H), $-$1.55 (s, 1H), $-$2.50 to $-$2.77 (m, 1H) (Fig. S5).

2.3. Preparation of PSN micelles

The solution of PSN in DMF (100$\mu$L) with a concentration of 10mg/mL was dropped into phosphate buffer solution (PBS) solution (2mL) with the aid of water bath ultrasound. Then, the mixed solution was concentrated via ultrafiltration, accompanied by DMF removal. The phthalocyanine concentration was determined through the standard curve for subsequent experiments.

2.4. Transmission electron microscope (TEM) characterization

A small amount of PSN suspension at different pHs was added onto a copper mesh (carbon supported film, 30mesh) and absorbed for 10min, followed by being negatively stained with 1% phosphotungstic acid for 10min. After natural drying at room temperature, the sample was exposed to TEM with an accelerating voltage of 100kV, to record its size and morphology.

2.5. Singlet oxygen yield test in aqueous solutions with different pH

Ethanol solution of DPBF and an aqueous solution of PSN at pH 6.5 or 7.4 were mixed in a cuvette. Then, the mixed solution was irradiated by laser for certain durations, and the absorption spectra were recorded to obtain changes in absorbance at 414nm with the ultraviolet spectrophotometer. The absorbance before illumination was regarded as the initial value $A_0$, and this value was changed to $A_t$ after different irradiation times. The curve that reflected singlet oxygen yield was plotted by taking time as the horizontal axis and $ln(A_0∕A_t)$ as the vertical axis. The larger slope of the curve represented the higher singlet oxygen yield.

2.6. Cell culture and cellular uptake experiment

4T1 murine breast cancer cells were cultured in DMEM including 10% FBS and 1% penicillin/streptomycin. In order to study cellular uptake, 4T1 cells were seeded on a 96-well plate at the density of $1\times 10^4$ per well and cultured overnight. The preseeded cells were then incubated with 1$\mu$M PSN for different times, followed by a fluorescence microscope to observe and capture intracellular fluorescence and the software ImageJ for quantitative analysis.

Then, the subcellular localization of PSN was investigated. 4T1 cells were seeded on 35mm dishes at the density of $2\times 10^4$ per dish and cultured overnight. Then, the preseeded cells were incubated with 1$\mu$M PSN for 4h and washed with PBS thrice. After being stained with Lysotracker for 30min and washed again, cells were fixed with 70% ethanol for 10min, followed by DAPI treatment for 15min and three-time PBS washing. Afterward, co-localization images were obtained with a confocal laser scanning microscopy (Nikon A1R high-speed confocal laser microscope, Nikon Corporation, Japan). By analyzing an arbitrary straight line crossing a certain cell in the picture using ImageJ, the spatial change was plotted to further quantify the co-localization of lysosome and PSN.

2.7. In vitro photocytotoxicity evaluation

To detect intracellular ROS, 4T1 cells were seeded on a 96-well plate at the density of $1\times 10^4$ per well and cultured overnight. The preseeded cells were then incubated with 1$\mu$M PSN for 4h, followed by incubation with 10$\mu$M DCFH-DA for 30min. Then, the cells were replaced with fresh DMEM and exposed to a xenon lamp for certain periods of time. The production of intracellular ROS was qualitatively and quantitatively analyzed with a fluorescence microscope and a microplate reader, respectively.

Then, the cells were seeded in the same way to study the photosensitivity of PSN. After incubation with PSN at different concentrations for 4h, the cells were replaced with fresh DMEM and subjected to dark treatment or laser irradiation treatment, followed by incubation with CCK-8for 1h. Then, absorbance at 450nm was measured by a microplate reader to calculate cell viability. To further realize the mechanism of cell death, 4T1 cells were seeded on a 6-well plate at the density of $5\times 10^5$ per well and cultured overnight. Then, the preseeded cells were incubated with 200nM PSN for 4h and replaced with fresh DMEM, followed by dark treatment or laser irradiation treatment. The apoptosis and necrosis rate of cells were measured using an apoptosis detection kit through the flow cytometer.

2.8. In vivo fluorescence imaging

Female Balb/c mice aged 6–8 weeks were purchased from BeijingVital River Laboratory Animal Technology Co., Ltd. All animal experiments were approved by the Peking University Institutional Animal Care and Use Committee (IACUC). An orthotopic breast cancer model was established by injecting $2\times 10^7$ 4T1-luci tumor cells into the fourth pair of mammary glands of each mouse. The volumes of tumors reached approximately 100mm³ about 6 days after injection and were used for fluorescence imaging in vivo. Then, PSN dissolved in PBS ($\mathrm{\text{pH}}=7.4$) was intravenously injected into the orthotopic tumor-bearing mice with a dose of 1$\mu$mol/kg. At different time points after the injection, a living animal fluorescence imaging system was used for fluorescence imaging with excitation at 660nm, and the fluorescence signal was collected using a 700nm filter. The data were analyzed by the FOBI software to get accumulation process of PSN at tumors.

2.9. In vivo photodynamic therapy

PDT was performed 7 days after the establishment of the orthotopic breast tumor model, and the average initial tumor volume was 100mm³. The mice were randomly divided into four groups ($n=5$): Group 1 with PBS, Group 2 with PSN, Group 3 with $\mathrm{\text{PM}}+\mathrm{\text{light}}$ and Group 4 with $\mathrm{\text{PSN}}+\mathrm{\text{light}}$. First, PSN (2$\mu$mol/kg), PM (2$\mu$mol/kg) and PBS, each of 200$\mu$L were intravenously injected to the corresponding groups. Tumor areas of Groups 3 and 4 were irradiated 4h later using a 671nm laser (0.2W/cm²) for 16min with a total laser intensity of 192J/cm². After treatment, the tumor size was measured for each mouse every day using the electronic vernier calipers for 15 days. The maximum width ($X$) and length ($Y$) of each tumor were measured, and the following formula was used to calculate the tumor volume: $V=(X^{2}Y)∕2$. The weight changes of each group were also recorded each day. Tumors in each group were randomly selected to be excised for H&E, TUNEL and Ki67 assay to analyze the tumor apoptosis, proliferation and necrosis. In order to investigate the biological safety of PDT with PSN, major organs (heart, liver, spleen, lung and kidney) of each group were also dissected for H&E staining to evaluate the pathological changes after the treatment.

2.10. Statistical analysis

Data were presented as $\mathrm{\text{mean}}\pm \mathrm{\text{standard}}$ deviation (SD). Student’s $t$-test was employed to test the significance of the difference, which was considered to be significant when *$P<0.05$.

3. Results and Discussion

3.1. Synthesis of PSN

A simple two-step synthesis method was used to synthesize asymmetric PSN in this paper (Fig. S1). Phthalocyanine silicon dichloride was synthesized by the reaction of isoindole and silicon tetrachloride in quinolone, followed by purification with precipitation and organic solvent washing. Then, phthalocyanine silicon dichloride and 1,3-bis(dimethylamino)-2-propanol underwent bilateral axial substitution reaction to obtain compound 1. After purification by basic alumina column chromatography, a unilateral nucleophilic reaction between compound 1 and PEG polymer with phenol active groups produced a PEG-phthalocyanine conjugate with pH response as the final product (compound 2, PSN). It is worth noting that the ratio of compound 1 and PEG polymer must be strictly controlled to minimize the formation of a by-product with bilateral nucleophilic substitution. PSN was purified with basic alumina column chromatography for subsequent experiments.

3.2. In vitro pH responsiveness of PSN

pH responsiveness is the key to reducing the phototoxicity of PSN to normal tissues. Various methods were used to investigate the responsive capacities of PSN to pH changes. Under a weakly alkaline condition ($\mathrm{\text{pH}}=7.4$), the inherent amphiphilic property promoted PSN to self-assemble into uniform micelles with mean diameters of less than 10nm, which has been confirmed by the TEM image and dynamic light scattering measurement. Whereas, when the solution became faintly acidic ($\mathrm{\text{pH}}=6.5$), protonation of dimethylamino turned PSN into hydrophilic molecules, thus resulting in the disintegration of micelles (Figs. 1(a) and S7). In order to observe the pH response characteristics of PSN more visually, PSN dissolved in solutions with two inverse pH were imaged by a small animal live imager. As shown in Fig. 1(b), the fluorescence intensity of PSN in a faintly acidic PBS is twice that of a weakly alkaline PBS. A more precise measurement was carried out via fluorescence spectroscopy, and the results displayed that faintly acidic conditions increased the fluorescence intensity by half [Fig. 1(c)]. This enhancement might be attributed to weakening of the PET effect after protonation rather than molecular disaggregation. The absorption spectrum also supported this inference because the absorption spectra of PSN in different pH PBS both presented a typical monomer state with a sharp absorption peak in the Q band [Fig. 1(d)], implying the dispersed state of PSN molecules in the micelles.

Compared with fluorescence, the ability to generate singlet oxygen is the crucial indicator for measuring the photodynamic killing effects of photosensitizers. It is necessary to investigate the effect of pH changes on the singlet oxygen yield of PSN. Singlet oxygen can oxidize DPBF with a characteristic absorption peak at 414nm into DBB without absorption, so absorption spectra that measured changes in absorbance at 414nm were performed to estimate the generation of singlet oxygen. As shown in Figs. 1(e) and 1(f), the photobleaching rate of PSN to DBPF was slow in the pH 7.4 PBS, but it was accelerated under faintly acidic conditions. The slope of the PSN singlet oxygen yield curve under faintly acidic conditions was 1.35 times that under weakly alkaline conditions (Fig. S9), indicating the pH-responsive singlet oxygen generation capability of PSN.

3.3. Cellular uptake and subcellular localization of PSN

Prior to evaluating the photodynamic effects of PSN in vitro, it is essential to investigate cellular uptake of PSN so that the optimal laser irradiation time can be determined. After incubating 4T1 cells with PSN for 2h, the red fluorescence of PSN was observed inside the cells by a fluorescence microscope. When the incubation time was extended to 4h, the intracellular fluorescence intensities reached the highest, which implied the maximum uptake of PSN by cells [Figs. 2(a) and 2(b)]. Hence, performing laser irradiation after incubation of PSN with tumor cells for 4h might be the optimal time for PDT in vitro. With the further extension of incubation time, the intracellular fluorescence gradually decreased until it almost disappeared at 24h. However, the positive charge after protonation endowed PSN with strong cell affinity followed by longer cell retention, so cells still maintained strong fluorescence after 10h of incubation.

Subcellular localization of PSN contributed to deeply realizing its intracellular behaviors. After incubation of PSN with 4T1 cells for 4h, confocal imaging was performed to observe the subcellular distribution of PSN. Lysotracker with green fluorescence and DAPI with blue fluorescence were selected to mark lysosomes and nucleus, respectively. As shown in Fig. 2(c), the red fluorescence of PSN co-localized with Lysotracker instead of DAPI, manifesting the concentrated distribution of PSN in lysosomes after the uptake into the cells. The plot profile analysis of a single cell through ImageJ also supported this conclusion [Fig. 2(d)]. These results indicated that PSN could be effectively taken up by tumor cells and accumulate in acidic lysosomes to restore its fluorescence.

3.4. pH-responsive phototoxicity of PSN against tumor cells

The ability of PSN to produce distinct intracellular singlet oxygen under different pH conditions is one of the keys to verify the pH-responsive phototoxicity of PSN against tumor cells. In order to simulate diverse environments of tumor cells and normal cells in vivo, cells were cultured in media with different pH ($\mathrm{\text{pH}}=6.5$ and $\mathrm{\text{pH}}=7.4$) in vitro. Afterward, PSN dissolved in different pH media was incubated with cells for 4h, succeeded by laser irradiation to generate singlet oxygen. DCFH-DA, a small molecule probe with free passage through the cell membrane, was employed for the detection of singlet oxygen. Singlet oxygen can oxidize nonfluorescent DCFH-DA into DCF with strong green fluorescence, allowing for qualitative and quantitative analysis of singlet oxygen production with fluorescence microscope and microplate reader. As shown in Fig. 3(a), PSN-incubated cell groups did not produce green fluorescence in the absence of laser irradiation, which was consistent with negative control groups incubated with PBS. Nevertheless, when exposed to laser irradiation, cell groups incubated with PSN under different pH conditions both exhibited green fluorescence, marking the generation of singlet oxygen. In particular, the intracellular fluorescence resulted from incubation at pH 6.5 was much stronger than that at pH 7.4, and further fluorescence quantitative analysis of DCF revealed that this amplification reached nearly fivefold [Fig. 3(b)], attesting that the acidic environment could stimulate PSN to produce a mass of intracellular singlet oxygen. These results fully declared the pH targeting properties of PSN, which meant that the photodynamic killing effect of PSN on tumor cells under acidic microenvironments was much stronger than that of normal cells even if the latter slightly took in PSN.

Based on the excellent pH responsiveness and singlet oxygen generation ability mentioned above, the cytotoxicity and photodynamic activity of PSN were evaluated by CCK-8 assay to provide safe and effective dose references for subsequent experiments. After incubating with PSN at different concentrations for 4h, cells were subjected to dark treatment or laser irradiation treatment. As shown in Fig. 3(c), when no laser irradiation was applied, PSN had no significant impact on cell viability even though its concentration was as high as 2$\mu$M, stating splendid safety and biocompatibility of PSN. However, under the motivation of light irradiation, PSN at a concentration of 25nM reduced the cell survival rate to less than 15%, and a concentration of 100nM or above killed almost all the cells. Such outstanding photodynamic activity of PSN might be ascribed to two aspects. On the one hand, the positive charge caused by protonation of pH-responsive groups promoted the uptake of PSN by cells. On the other hand, the nonaggregated state as a result of axial modifications together with the reduction of the PET effect between PSN molecules under an acidic environment both enhanced the singlet oxygen yield of PSN. To further realize the damage mechanism of PSN to the cells, flow cytometry was utilized to analyze cell status after PDT. As shown in Fig. 3(d), the apoptotic rate of cells after PDT was as high as 38.9%, suggesting that the cell death is dominated by cell apoptosis induced by singlet oxygen.

3.5. Tumor-specific accumulation of PSN

Through monitoring tumor-bearing mice with a living animal fluorescence imaging system, the enrichment of PSN in tumor sites and its pH sensitivity in living animals were investigated as a prerequisite for the subsequent PDT in vivo. As shown in Figs. 4(a) and 4(b), tumor sites produced strong fluorescent signals 5min after intravenous injection of PSN micelles dissolved in PBS. As time went by, fluorescence signals of tumor sites continued to increase until they reached peak values at 3h after injection, hinting at 3h as the best candidate for light irradiation time of PDT in vivo. Whereafter, tumor tissues maintained strong fluorescent signals up to 60% of the peak signals within 72h after injection. The rapid appearance and long-term retention of fluorescent signals at tumor sites were roughly attributed to three points. First, PSN micelles could hold their nanostructures when circulating in a weakly alkaline blood physiological environment, allowing for their fast accumulation and penetration into tumor tissues through the EPR effect. Once entering tumor sites, protonation of PSN caused by the acidic TME produced strong fluorescence in tumor tissues to realize a high signal ratio to surrounding normal tissues. Lastly, positive charges after protonation enabled PSN to be easily taken in by tumor cells and retained for a long time.

Imaging results of isolated organs further illustrated the tumor specificity of PSN caused by pH response. The mice were sacrificed 24h after injections of PSN, and their major organs and tumors were dissected for fluorescence imaging. As shown in Figs. 4(c) and 4(d), tumor tissues presented strong fluorescence signals while fluorescent signals of other organs such as liver and kidney were pretty weak. These results indicated that PSN could enrich and restore photosensitivity at tumor sites through the EPR effect and pH response, thus leading to powerful photodynamic damage to tumor tissues. When it came to normal organs, PSN would maintain their nanostructures to weaken singlet oxygen generation ability so that side effects to normal tissues could be minimized.

3.6. Selective PDT of PSN against tumor

Based on the splendid tumor specificity of PSN, the in vivo PDT potential of PSN was investigated. In order to better compare and verify the PDT enhancement effect induced by the pH response of PSN, micelles composed of non-pH-responsive phthalocyanine (PM) and DSPE-PEG2000 were set up as a control group. In contrast to PSN micelles, PM micelles were physically embedded and insensitive to pH, so their differences were able to verify the influence of the embedding form and pH response. As shown in Fig. 5(a), tumors treated with $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ were completely ablated without recurrence 15 days after treatment, while the tumor volume of the group treated with $\mathrm{\text{PM}}+\mathrm{\text{light}}$ only decreased in the first 3 days of treatment but gradually increased at a rate comparable to that of the PBS-treated control group later on. Such obvious differences in therapeutic efficacy confirmed the splendid PDT potential of PSN, which might be owed to three factors. First of all, the $\mathrm{\text{PM}}+\mathrm{\text{light}}$ group had a potential defect of photosensitizer leakage during blood circulation due to simple physical embedding, so its photosensitizers were less enriched at tumor sites than the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group. Next, photosensitizers in the $\mathrm{\text{PM}}+\mathrm{\text{light}}$ group could not be protonated into a nonaggregated state conducive to PDT because of a lacking pH response. Third, negative charges of the carboxyl group prevented photosensitizers in the $\mathrm{\text{PM}}+\mathrm{\text{light}}$ group from being effectively taken up by tumor cells.

To further reveal anti-tumor mechanisms of PSN-based PDT, tumor tissues after treatment were sliced for H&E, TUNEL and Ki67 staining [Fig. 5(b)]. As shown by TUNEL staining, no green fluorescence was observed in tumor slices of the PBS and the PSN group. In contrary, tumor slices of the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group exhibited the strongest green fluorescence, indicating the highest degree of apoptosis. Subsequent fluorescence quantitative results showed that the number of apoptosis in the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group was twice that in the $\mathrm{\text{PM}}+\mathrm{\text{light}}$ group [Fig. 5(c)], so the former achieved a far better tumor suppressing effect than the latter. This apoptosis difference might be attributed to the different enrichment and cellular uptake at tumor sites as a result of disparate embedding forms and charges. Through Ki67 immunohistochemical analysis, it was observed that Ki67 staining in the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group was the least, only 1/10 of the PBS control group [Fig. 5(d)], which revealed significantly attenuated cell proliferation ability of the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group due to the strong anti-tumor effect of selective PDT. H&E staining also intuitively supported the strongest killing effect of the $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ group on tumors. These results all proved the superiority of pH-responsive PSN micelles for selective PDT.

Finally, the safety of PSN for PDT in vivo was discussed. Mouse body weight with various treatments was recorded over time. No obvious body weight change was observed among all groups (Fig. S10a), suggesting good safety of PSN-based PDT. Moreover, H&E staining pathological examination also verified that the main organs of mice treated with $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ were intact (Fig. S10b). Therefore, PSN is a photosensitizer with good safety, strong tumor selectivity and splendid photodynamic efficacy, which gives it a broad application prospect for selective PDT against tumor.

4. Conclusion

In summary, we have successfully designed and synthesized asymmetric PSN with pH-responsive function for selective PDT against tumors, with minimum damage to normal tissues. PSN could self-assemble into micelles under weakly alkaline conditions and exhibited feeble fluorescence and singlet oxygen generation ability on account of the PET effect. Once the environment became faintly acidic, methylamino groups of PSN were protonated, followed by disintegration of micelles together with recovery of fluorescence and singlet oxygen yield due to weakening of the PET effect. Experimental results in vivo declared that PSN accurately targeted tumors and retained them for a long time. Tumors of mice treated with $\mathrm{\text{PSN}}+\mathrm{\text{light}}$ were completely eliminated without recurrence, which was far greater than ordinary phthalocyanine without pH response. We believe that PSN as a pH-responsive photosensitizer will have broad application prospects for selective PDT against tumor and facilitate remarkable advances in cancer therapy.

Conflict of Interest

The authors declare that there are no conflicts of interest relevant to this article.

Acknowledgments

This work was supported by grants from projects of Interdisciplinary Research Foundation of HIT, the National Natural Science Foundation of China (No. 82071980), the International Cooperation and Exchanges NSFC-PSF (No. 31961143003), the State Key Program of National Natural Science of China (No. 81930047) and the National Project for Research and Development of Major Scientifc Instruments (No. 81727803). Y. Zhou, W. Zeng and M. Wang contributed equally to this work.